Verifiable work & Identity in the coming era
I was thinking this morning about how the progression of AI as being "good enough" is going to force a change on homework that is AI doable. Math, writing, computer science, all of these areas are ones we're seeing AI become capable enough to solve in a way which would make it hard to identify whether the student did it on their own, or with aid.
I noted two posts by a friend of mine on Mastodon in yesterday's favs, included again below:
On the one hand, I believe many common examples of homework to be remote busywork. At best, misguided as to the efficacy, and at worst, just work to be able to feel the student put in effort. Yes, there were definitely examples that I felt helped me understand and grow in my abilities in different academic areas, but they were exceptions. As a layman with no evidence to back this up, my feeling is that there is a sizable dip in effectiveness where the earliest years of schooling benefit from the added practice, and that in the later years of schooling it has similar levels of benefit as projects become more complex and require more than the time allotted in school. Additionally, team projects which require collaboration were good practice for communicating with others, rudimentary project management concepts, and also understanding how others work (or often didn't work.)
As Adam noted, there will likely be a growing importance in these sorts of projects to provide additional resources, proof of work, research, notes, as means of helping show your process and work, and as another piece of verification for work completed by you and not an AI assistant.
Additionally, in this same vein of verification against AI, comes the concerns over deepfakes, or more subtly the impersonation of someone in text form on social media or online. How do we know our friend really said those things when AI can generate a good-enough doppelgänger in many cases?
One idea I saw floated, though I didn't flag it so it is now lost to me, was that online identity verification could be an example of a blockchain style public ledger for verification. Obviously it has numerous security, privacy, and technology concerns; but it did strike me as perhaps the most compelling use-case for blockchain style technology that I have heard in a while. I recognize that I am not well informed enough to know if it is indeed viable; but it at least got my gears going.
Anyways, AI and this new era is definitely on my mind and likely will continue to be.